Summary

  • Before DNS, JUNET used hierarchical mail names; local mail configuration and UUCP gateways made them useful across a growing research network.
  • When JUNET moved from .junet to .jp in 1989, most organization-level names could be carried forward. The volunteer junet-admin group continued assigning names until JNIC took over the wider .jp registry in 1991.

Analysis

An early JUNET mail name looks familiar to anyone who has used email. Its form suggests a hierarchy of user, host, organization and suffix. But in mid-1980s Japan, that hierarchy did not imply that messages traveled over the Domain Name System or the Internet Protocol.

JUNET began in October 1984 by connecting UNIX computers at the University of Tokyo, Tokyo Institute of Technology and Keio University over telephone lines. The network used UUCP, a way for UNIX systems to exchange files and messages over intermittent links. Email and electronic news were not future applications waiting for a global backbone; they were the working reason to connect the machines. The first JUNET User’s Guide describes a network assembled from participating organizations’ local networks, with gateways bridging differences and UUCP links running over public telephone lines or packet networks.

The arrangement grew beyond three campuses. By November 1987, the guide counted about 110 participating organizations and roughly 1,000 systems. Those numbers do not describe one centrally managed machine room. A participating institution operated its own local network and gateway; messages crossed the links that connected those gateways. The system needed names that users could write and administrators could translate into the delivery rules each host actually used.

A naming tool called mailconf, developed in 1985, generated sendmail configuration from mail-delivery rules. JPNIC’s history cites a hierarchical JUNET address that the tool made usable. The form let a reader recognize the host and its organization even though DNS was not resolving the name. UUCP supplied the transport; local mail configuration and gateway knowledge supplied the path. The visible name was not itself a route.

The volunteer junet-admin group managed the uniqueness of organization-level labels. In the example, titech identified Tokyo Institute of Technology; the group did not need to assign a unique name to every device on every campus. This was a practical division of labour: institutions administered their own machines, while a shared group prevented the organizational names from colliding inside JUNET. The name hierarchy was therefore a coordination tool fitted to the network that existed, not a miniature copy of a global DNS registry.

International exchange widened the stakes. JUNET connected to the US Computer Science Network in January 1986, and JPNIC says email and news exchange with overseas networks became routine. Other Japanese networks—including BITNETJP and HEPnet-J—also had their own naming practices. A label useful inside one experimental network was beginning to matter across a larger set of institutions and connections.

The move to .jp was prepared before the old suffix disappeared. JPNIC’s history says Jon Postel delegated .jp to Jun Murai in 1986; in 1988, the major academic networks endorsed the delegation and a shift to .jp. The new namespace introduced organization-type labels such as AC, CO and OR. The design had an operational edge: academic-network acceptable-use rules distinguished research traffic from other uses, and network operators explored using organization types to guide access across paths. The record describes an intended control surface, not evidence that every operator enforced the same rule in the same way.

In April 1989, JUNET names moved from .JUNET to .JP. JPNIC says that, with some exceptions, the part to the left of the suffix was retained while .JUNET was replaced with an organization-typed ending such as .AC.JP or .CO.JP. The transition worked because an organization already had a managed label that could survive a change in the top of the tree. That is an inference from the documented mapping, not proof that each mailbox migrated without error. The available histories do not publish a complete before-and-after host list, bounce rate or failure count.

Changing the suffix did not instantly create a national registry. No dedicated organization yet existed to assign .jp names, so junet-admin continued doing that work for networks beyond JUNET. In December 1991, the Japan Network Information Center (JNIC) took over. JPNIC’s history says networks wanted an organization they could share, one that did not belong to any single network. The near-contemporary JPNIC newsletter describes the volunteer group’s continuing work after the migration; the later account explains why the expanding namespace called for a broader information centre.

JUNET itself remained an experimental network, eventually stopping in October 1994 as other projects built IP-based connectivity. Its contribution was not that it secretly ran DNS early. It was that a community could make names useful before DNS by combining conventions, software, volunteer assignment and working gateways. The .junet suffix then became replaceable because the organization-level structure beneath it was already understood. What had to change next was the scope of stewardship: a name service used by several networks could no longer be treated as a task belonging only to one of them.

Sources